Views
No views yet
README.md I provided earlier. It’ll make your model page much easier to understand and more shareable.1---
2license: apache-2.0
3tags:
4- gpt-neox
5- causal-lm
6- lora
7- fine-tuned
8- text-generation
9- huggingface
10model-index:
11- name: finetune-DO
12 results: []
13---
14
15# 🚀 Fine-tuned GPT-NeoX-20B on English Quotes
16
17This model is a fine-tuned version of [EleutherAI/gpt-neox-20b](https://huggingface.co/EleutherAI/gpt-neox-20b) using LoRA (Low-Rank Adaptation) on a quotes dataset from [Abirate/english_quotes](https://huggingface.co/datasets/Abirate/english_quotes).
18
19Fine-tuning was performed by [@Dwayne234](https://huggingface.co/Dwayne234) using Hugging Face 🤗 and Google Colab with 4-bit quantization enabled (BitsAndBytes).
20
21---
22
23## 🧠 Model Details
24
25- **Base Model**: GPT-NeoX-20B
26- **Adapter Type**: LoRA
27- **Quantization**: 4-bit (NF4, bfloat16 compute)
28- **Training Steps**: 10
29- **Frameworks**: 🤗 Transformers, PEFT, BitsAndBytes
30
31---
32
33## 📦 How to Use
34
35```python
36from transformers import AutoModelForCausalLM, AutoTokenizer
37
38model = AutoModelForCausalLM.from_pretrained("Dwayne234/finetune-DO")
39tokenizer = AutoTokenizer.from_pretrained("Dwayne234/finetune-DO")
40
41prompt = "Ask not what your country"
42inputs = tokenizer(prompt, return_tensors="pt")
43outputs = model.generate(**inputs, max_new_tokens=20)
44print(tokenizer.decode(outputs[0], skip_special_tokens=True))| Setting | Value |
|---|---|
per_device_train_batch_size | 1 |
gradient_accumulation_steps | 4 |
max_steps | 10 |
learning_rate | 2e-4 |
fp16 | True |
optimizer | paged_adamw_8bit |
</details>
3. Add a commit message like `Update model card with full README`.
4. Hit **Commit changes**.
---
Once saved, refresh your model page — it'll now look clean and informative.
Want me to add a badge or preview your model card before publishing?